Online Analytical Processing

Two colleagues analyzing data on a laptop, illustrating the concept of Online Analytical Processing.

Online Analytical Processing (OLAP), as opposed to Online Transaction Processing (OLTP), supports data analysis using a live connection to structured data to reveal analytic insights. OLTP systems create data, and OLAP systems analyze business data.

Why is Online Analytical Processing Important?

Businesses must act quickly to stay competitive, respond to threats and opportunities, and respond to customer needs. Online analytical processing provides visibility into these changes in the business environment.

The Evolution of Online Analytical Processing

Transaction processing systems are architected for high throughput, large user populations, and high-speed data creation. For example, Banking systems such as automatic teller machines (ATMs) and point of sale (POS) systems in supermarkets must be highly responsive. Early reporting systems would gather transaction data at night when the transaction systems provide batch reports to be consumed the following day.

Reporting systems evolved into data warehouses and OLAP cube-based systems that rolled up data to allow for multidimensional analysis of transaction data. The problem with early OLAP systems was that the data cubes or hypercubes they created lacked access to the most current underlying data.

Today’s data warehouse systems can be coupled to message-based streaming data sources to gather data from transactional systems in sub-seconds after creation. Machine Learning (ML) systems can detect subtle trends and correlations in raw data streams that can be displayed in dynamic visualizations. Real-time dashboards in Business Intelligence systems enable organizations to respond to changes in the moment.

Hybrid OLTP and OLAP Systems

To minimize the latency between transaction system data and data analysis, products such as Actian Ingres have emerged to address this need. In this hybrid system, the Ingres OLTP database engine is focused on transactional workloads coupled with the Vector analytic database, which stores data analytic data. It is a single database instance that uses row storage for OLTP tables and columnar storage for tables for decision support. A keyword in the CREATE TABLE tells the Actian X database the table’s intended use so it can be optimized for OLTP or OLAP data.

OLAP cubes

A class of databases that preloads data into a multidimensional cube with pre-aggregated data to support slicing and dicing data across different dimensions. These databases use a non-standard multidimensional expressions (MDX) query language. OLAP cubes have been largely displaced by columnar database technology, which uses standard SQL queries and can refresh data in real time.

The Benefits of Online Analytical Processing

Users of decision support applications are increasingly demanding fresher data for their analysis. Below are many of the benefits of using the latest available analytic data:

  • When a business becomes aware of changes in its environment, it often needs to respond quickly to minimize damage to its reputation. Monitoring customer feedback on new products and services requires adapting to negative comments or changes in the sentiment of social media comments about the company or product before they escalate.
  • Fraud detection systems need the latest information before a business makes a loan decision or prices risk into an insurance premium.
  • As market opportunities are created, such as a heatwave in a particular region, a hardware retailer needs to stock up on fans and air conditioning systems while the heat lasts.
  • Changes in supplier pricing dictate the price a manufacturer charges their customers. The sooner they respond, the more likely they are to avert a drop in profit margins.

Actian and Online Analytical Processing

The Actian Data Platform provides a unified data management system that includes the Actian Vector data analytic database and DataConnect to integrate operational data from its source system into an analytics-ready state. You can try the Actian Data Platform using a free trial here.